Proposed in this paper is a discrete particle swarm optimization model to solve set-based combinatorial optimization problems.The model introduces set concepts and operations in the particle swarm optimization,defines a search space of variable set and redefines the velocity and location of particles as well as the operators working in the defined search space.Thus,it possesses the characteristics of both particle swarm optimization and set-based combinatorial optimization.Finally,the proposed model is applied to the knapsack problem,a typical set-based combinatorial optimization problem,and it is compared with the binary particle swarm optimization（BPSO）.The results indicate that the proposed model is of stronger searching ability and higher stability.